Task-Core Memory Management and Consolidation for Long-term Continual Learning

2025-05-15

AI Summary

This paper addresses the exacerbated catastrophic forgetting in long-term continual learning (Long-CL) under massive, streaming task sequences. Inspired by human memory mechanisms, we propose a novel framework featuring a task-centric memory management strategy and a long-term memory consolidation mechanism—enabling adaptive indexing of critical samples and selective retention of hard-to-discriminate ones—integrated with memory indexing, dynamic updating, and selective consolidation. We systematically evaluate our approach on two self-constructed ultra-long-term benchmarks: MMLongCL-Bench (multimodal) and TextLongCL-Bench (text-only), achieving absolute improvements of 7.4% and 6.5% in average accuracy (AP), respectively, surpassing state-of-the-art methods. Our core contribution is the first formal incorporation of long-term memory consolidation modeling into continual learning, enabling synergistic optimization of knowledge acquisition efficiency and long-horizon retention capability.

Technology Category

• Machine Learning: Life-Long and Continual Learning
• Search and Optimization: Learning to Search
• Natural Language Processing: Learning & Optimization for NLP

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Studies of user behavior, including longitudinal effects of personalized systems

Abstract

In this paper, we focus on a long-term continual learning (CL) task, where a model learns sequentially from a stream of vast tasks over time, acquiring new knowledge while retaining previously learned information in a manner akin to human learning. Unlike traditional CL settings, long-term CL involves handling a significantly larger number of tasks, which exacerbates the issue of catastrophic forgetting. Our work seeks to address two critical questions: 1) How do existing CL methods perform in the context of long-term CL? and 2) How can we mitigate the catastrophic forgetting that arises from prolonged sequential updates? To tackle these challenges, we propose a novel framework inspired by human memory mechanisms for long-term continual learning (Long-CL). Specifically, we introduce a task-core memory management strategy to efficiently index crucial memories and adaptively update them as learning progresses. Additionally, we develop a long-term memory consolidation mechanism that selectively retains hard and discriminative samples, ensuring robust knowledge retention. To facilitate research in this area, we construct and release two multi-modal and textual benchmarks, MMLongCL-Bench and TextLongCL-Bench, providing a valuable resource for evaluating long-term CL approaches. Experimental results show that Long-CL outperforms the previous state-of-the-art by 7.4% and 6.5% AP on the two benchmarks, respectively, demonstrating the effectiveness of our approach.

Problem

• Evaluating existing continual learning methods in long-term scenarios
• Mitigating catastrophic forgetting during prolonged sequential updates
• Developing memory management and consolidation for robust knowledge retention

Innovation

• Task-core memory management for efficient indexing
• Adaptive memory updates during learning progress
• Selective retention of hard discriminative samples